THE QUANTITATIVE ESTIMATION OF TOTAL IRON STORES IN HUMAN BONE MARROW

THE QUANTITATIVE ESTIMATION OF TOTAL

IRON STORES IN HUMAN BONE MARROW

Edward Gale, … , John Torrance, Thomas Bothwell

J Clin Invest.

1963;

42(7)

:1076-1082.

https://doi.org/10.1172/JCI104793

.

Research Article

Find the latest version:

Vol. 42, No. 7, 1963

THE

QUANTITATIVE

ESTIMATION OF TOTAL IRON STORES

IN

HUMAN BONE

MARROW

By EDWARD GALE, JOHN TORRANCE, AND THOMAS BOTHWELL * (From the Department of Medicine, University of Witwatersrand Medical School,

Johannesburg, South Africa)

(Submitted for publication December 3, 1962; accepted March 7, 1963)

In

normal adult

males,

about

25%o

of the

body

iron

content

is in storage

depots (1).

This iron

exists in

two

forms:

as a

diffuse

soluble fraction

called

ferritin,

in which the molecules

are

dis-persed,

and

as

insoluble aggregates

of

hemosiderin,

which

can

be

visualized

by

conventional

micros-copy

(2).

Although

the liver is

regarded

as

the

chief storage organ,

chemical

analyses suggest

that

it

normally

contains up

to

300

mg

(3-5),

which

is

only

between one-quarter

and one-third

of

what

can

be mobilized

from total

stores

when

healthy young males

are

repeatedly phlebotomized

(6).

Although

little is known

of

the

quantities

present in other organs, hemosiderin

can

be

seen

in

the reticuloendothelial cells of the bone

mar-row

(7)

and

spleen

(8),

and

there is

some

chemi-cal evidence

to

indicate that

significant

amounts

of storage iron may be present in skeletal muscles

(9,

10).

The

present

study was undertaken

to

find out

how much iron is

normally stored in the

reticulo-endothelial cells of the bone

marrow

and

to

de-fine

the extent to which these stores

can

expand

when iron

overload

is

present.

In

addition,

a

comparison

was

made between the concentrations

of

iron present in the bone marrows and livers of

subjects with

varying

quantities of storage

iron.

MATERIALS AND METHODS

Clinical material. The total iron stores present in bone marrow were estimated in 61 adult subjects un-dergoing thoracotomy. Forty-one were white and the remainder were Bantu. The individual diagnoses are shown in Table I.

The chemical concentrations of storage iron were

esti-mated in the livers and bone marrows of adult Bantu and white subjects dying in the hospital. The Bantu

*_{This work was supported by grant AM04912-02}

from the National Institutes of Health and by a grant from the Council for Scientific and Industrial Research,

South Africa.

specimens (58 males and 44 females) were obtained from Baragwanath Hospital, Johannesburg, and the white specimens (25 males and 28 females) from the

General Hospital, Johannesburg.

The use of radioiron as a marrow label. When a tracer quantity of radioiron is injected intravenously,

most of it is taken up by the red-cell precursors of the bone marrow (11). This iron is subsequently

re-leased over the next few days as part of the hemo-globin of new redcells, and the percentage present in

cir-culation at 10 to 14 days has been used as a measure of the fraction initially taken up by the bone marrow (12).

If a specimen of bone marrow is therefore removed

be-tween 18and 24 hours after the injection of the radioiron (i.e., at a time when marrow activity is maximal, and

circulating activity is negligible), it is possible to use the Fe' as a label for relating values in the sample to the total marrow. This principle, which has previously been employed for estimating the total number of cell precursors (13-15), was used in the present study as a

means of calculating the total amount of storage iron present in the bone marrow. To do this, we must

as-sume that the reticulum cells containing storage iron are

fairly evenly dispersed throughout the erythroid marrow. Isotopic techniques. Subjects undergoing thoracotomy

weregiven between 5 and 10

_/Ac

high specificactivity Fe"9 citrate intravenously between 18 and 24 hours before surgery. A weighed sample of this solution was set aside as a standard. Blood samples were collected over the next few hours, and plasma volumes were then cal-culated from the line of clearance of radioiron as de-scribed previously (16). The total blood volume was

obtained from the plasma volume and packed cell

vol-ume corrected for the body:venous hematocrit ratio at an altitude of

5,900

feet

(17).

With this approach, the

mean figure for blood volume in the group was 68.2 ml per kg (SD,

10.7).

The rib removed at operation was squeezed dry, and the marrow so obtained was weighed and counted for radioactivity in a well-type counter. The amount of storage iron present was then measured chemically (see Chemical

methods).'

Blood samples were collected from all the subjects 10

to 14 days after theinjection ofradioiron. The total per-centage of injected radioactivity in circulation was then calculated by using the estimation of blood volume ob-tained from radioiron data on the first day of the study. The meanfigurefor the group was 83% (SD, 10.7). On

TOTAL IRON STORES IN HUMAN BONE MARROW

the assumption that the radioiron present in circulation at this time was in the marrow 18 to 24hours after the original injection, it was then possible to obtain a

meas-ure of the total iron stores present in bone marrow.

Marrow iron stores= (fraction of injected Fe' in cir-culation at 10 to 14 daysXcounts injected/counts in sample) X nonheme iron in marrow sample.

Chemnical Methods. The quantity of nonheme iron present in specimens of marrow was estimated by a

modification of the method of Brilckmann and Zondek (18). Four ml saturated sodium pyrophosphate was added to the marrow and sufficient 20%o trichloroacetic acid to giveafinal volume of 10 ml. (Inthis calculation

thespecific gravityof themarrow was assumedtobe1.0).

After thorough mixing, the tube was heated in a water bath and was maintained at a temperature of 800 C for 10 minutes. During this time the solution was stirred frequently. After cooling, the mixture was filtered and

an iron estimation was carried out on a sample of the filtrate by use of the thioglycolic acid method (19).

The validity of the whole procedure as a means of separating heme from nonheme iron was established in preliminary studies with solutions of hemoglobin and ferritin and suspensions of hemosiderin granules.

Fer-so

(A UJ

w 0

CA z

t

ac

_E

L._

_

._

Je 6

-i

1a

1

WHITE WHITE

FEMALES MALES

TABLE I

Clinicaldiagnosesin 61 patients

subjected

to

_thoracotomy

White Bantu

Diagnosis M F M F

Mitralstenosis 7 9 0 0

Intrathoracic neoplasms 8 0 1 0

Pulmonary tuberculosis 3 0 8 3

Pulmonary sepsis 0 4 6 2

Hiatus hernia 6 4 0 0

Total 24 17 15 5

ritin was prepared by the method of Mazur and Shorr (20), and hemosiderin by the method of Shoden and Sturgeon (21). Only 5% (range, 3 to8%) of heme iron

was present in the filtrate. Calculations based on the quantities of hemoglobin present in marrow samples in-dicated that theerror that could be introduced by hemo-globin iron was not more than 10 ,Ag per g marrow. The mean recovery of ferritin iron in concentrations

varying between 100 and 2,000

jug

per ml was 104%o

BANTU FEMALES

BANTU MALES

FIG. 1. CALCULATED VALUES FOR TOTAL MARROW IRON STORES IN 61 HUMAN SUBJECTS.

1077

)OO

_{_I}

U

*

No

I

0I

0 U

0

oo

~~~

I

o

D80o

1

00~~~~~~

~ 0

00~~~~~~

0 ~ ~ C3

0

0~~~ 0

100 ~ ~ 0

80

IG~~~

10000

E 1000

z

I-0% Z °

O w

I- E

z

LU L U C"

u

100

z

0

0 WHITE FEMALES ,

0 _{WHITE MALES} *

* BANTU FEMALES * .

* BANTU MALES .

.

*/*//#

.

0

~

*/ _f O

amsye.

°~~~~**U,~ ~ o a

o

0 co

0

Osa

- ,

0 0

0 0~~

0

0o

T I III I I1I1I1II

10 100 1000 10000

IRON CONCENTRATION IN MARROW

micrograms per gram

FIG. 2. CORRELATION BETWEEN STORAGE IRON CONCENTRATIONS IN THE MARROWS AND LIVERS OF

155 NECROPSY SUBJECTS. The regression line is indicated (r=+0.88, p<.001).

(range 100 to 106%), whereas that of hemosiderin iron

in concentrations varying between 70 and 10,000 jug per ml was 103% (range97 to 111%).

In necropsy studies, specimens of rib and samples of

liver were collected. Marrow was obtained from the

ribs by squeezing them dry, and the concentrations of

nonheme iron were estimated with the method described

in the previous paragraphs. As the validity of results

obtained for the concentrations of storage iron in mar-row depended on there being a moderately constant

re-lationship between storage iron and the weight of mar-row samples, some preliminary studies were done in which the concentrations of storage iron were estimated

in samples of marrow obtained from different sites. Specimens were taken from two sites in subjects with

concentrations of marrow storage iron varying between

15 to 4,476 ug per g. The average difference from the mean of each pair ofobservations was 5.9% (range 0 to

23%) in 16cases. It was felt that this degree of

agree-ment was sufficiently close for meaningful information

to be obtained from single samples.

The concentration of total iron present in each forma-lin-treatedsampleofliver wasestimated asdescribed

pre-viously (8). In addition, the concentration of heme iron

was estimated on a separate sample of liver, and the

concentration of storage iron was then obtained by

sub-tracting this result from the figure for total iron. This proved somewhat difficult as there are no simple, de-scribed techniques for the quantitative estimation of

he-moglobin iron in formalin-treated tissues. The method

eventually devised was as follows. Approximately 1 g

formalin-treated liverwasblotted and weighed. The

sam-ple was then cut into slices and placed in a 50-ml

volu-metric flask. One-half ml of the detergent sodium alkyl sulfate1 was added, and the flaskwasmade uptovolume with 1 N sodium hydroxide. The addition of the

deter-gent ensured a clear final filtrate. The flask was placed in an oven at 560 C for 18 hours. After cooling, the

solution was filtered, and the optical density of the

al-kali-heme wasdetermined by using anEvelyncolorimeter

with a 540 filter. The concentration of hemoglobin iron

present was then read from a calibration curve con-structed with hemoglobin solutions of known iron

con-1_Marketedas Teepol by the Shell Chemical Co.,

TOTAL IRON STORES IN HUMAN BONE MARROW

tent which had been treated in the same way. With

this technique, reproducible calibration curves were

ob-tained on solutions of hemoglobin that had been stored

in formalin for as long as 3 months.

Inan attempt to establish the specificity of the method,

estimations of hemoglobin iron were carried out on

deeply jaundiced livers obtained from rats in which the

bile duct had been tied several days previously. The

results wereno higher than in normal rats. In addition, the fact that concentrations of heme iron were in the same range as those obtained by other methods on

fresh liver specimens (9, 22) further supports the validity

of the present technique.

Histological methods. In the majority of studies, samples of marrow specimens were squashed onto slides

that were then stained for iron by Dry's method (23). The arbitrary histological gradings of storage iron in marrow particles were as follows: grade 0-no visible iron under oil immersion (magnificationX720); grade

1+-small iron particles just visible in reticulum cells

under oil immersion (magnificationX720); grade

2+-°

E

1000

a' Om

Z 0

I.) O E < 0o 6

wJ E 100

small, sparsely distributed iron particles usually visible

under low power (magnificationX80); grade 3+-nu-merous smallparticles presentin reticulum cells

through-out the marrow particles; grade 4+-larger particles

throughout the marow with tendency to aggregate into clumps; grade 5+-dense, large clumps of iron through-out the marrow; and grade 6+-very large deposits of

iron, both intra- and extracellular, which obscure

cellu-lar detail in the marrow particles.

RESULTS

Total iron

stores

in the

marrow

(Figure 1).

The

mean

value in 17 white females

was

99

mg

(range 19

to

237 mg), and

in

24 white males,

288

mg

(range 10

to

833 mg).

Only

3 of

the

males had values below 100

mg,

and

2 of

these

subjects

gave a

history of previous

gastrointestinal

hemorrhage. The 3 males with figures above 500

mg were

all

suffering from bronchial carcinoma.

IV

0 1+ 2+ 3+ 4+ S+ 6+

HISTOLOGICAL GRADING OF MARROW IRON

FIG. 3. RANGE OF CHEMICAL CONCENTRATIONS OF STORAGE IRON IN THE MARROWS OF 199

SUB-JECTSIN WHOM THE EXTENTOFRETICULOENDOTHELIAL STORES WAS ASSESSED HISTOLOGICALLY.

a I I I - z

*la

000

~~ ~ ~ ~ ~

::

*.

000

0*

~

~

~

~

~~0

o*

0~ ~ ~ ~ ~ ~ ~~ 0

~

*::

0

.0

*

*.:

00*

* 000

*~~~~ ~

.0Ye

040

14-4

~

~

X~ i

In

I

1. I

Only

5

Bantu females

were studied. Results

showed

a

wide

scatter,

with values

ranging

be-tween

56 and

781

mg.

The

mean

value in 15

Bantu

males

was

1,629

mg

(range

304

to

4,820

mg)

Correlation between concentrations of

storage

iron in

marrow

and

liver (Figure 2).

The

re-sults in

155

subjects

show

that the concentrations

of

storage

iron

present

in

the

twoorgans were

ap-proximately the

same over a very

wide

range.

Correlation

between

chemical

and

histological

assessment

of

storage

iron

in

the

marrow.

The

re-sults in 199

subjects

are

shown in

Figure

3.

Al-though

there

was some

degree

of

overlap,

moder-ately good

agreement was

obtained between

the

histological gradings

and

the chemical

concentra-tions of

storage

iron

in

the

marrow.

The

mean

figures

for each

grade

were as

follows: 0

=

43

_ttg per g

(SD, 23); 1+

=

130

jug

per g

(SD, 50);

2+ =223

ugper g

(SD, 75); 3+

=

406

ptg per g

(SD, 131); 4+

=

762

ug per g

(SD, 247); 5+

=

1,618

_/.tg

per g

(SD, 464); and 6+ = 3,681 ,ug

perg

(SD, 1,400).

DISCUSSION

Estimates of the

amounts of iron

present

in

stores

have

been

made in different ways. When

normal

subjects

are

repeatedly phlebotomized,

the

red-cell

deficit is corrected

by mobilizing

iron

from

stores.

Calculations

from such data

indi-cate

that the total

stores in

healthy

young males

vary

between

1,000

and

1,500

mg

(6). Although

data derived

from

an

isotopic

dilution

technique

have

given

a mean

figure

of

only

600 mg in man

(24), these discrepancies

are not

surprising,

since

the

response to

repeated phlebotomies

is a meas-ure

of the total

mobilizable stores, whereas the

isotopic technique

is

only

a measure of miscible

stores.

The

major

sites of

storage

iron

in the

body

have

been

demonstrated

by using

both

histological

and

chemical

methods.

The liver

normally contains

up to

300

mg, and

significant quantities

are

pres-ent

in the

reticuloendothelial cells of the bone

marrow

(7, 25)

and

spleen (3, 8).

In

addition,

there is

some

chemical evidence

to suggest

that

significant

amounts

of

storage

iron

may

be

pres-ent

in

skeletal muscles

(9, 10).

There

is,

how-ever,

little

quantitative information

on

the total

amounts

of iron

present in organs other than the

liver

and spleen.

In

the present study,

an

attempt

was

made

to

estimate the

total

amount of

storage iron in the

bone marrow.

This

was

done

by

adapting

an

isotopic

dilution technique which has been

previ-ously used for gauging the number of cell

pre-cursors

in bone

marrow

(13-15).

The data

ob-tained with this

technique

indicate

a mean

figure

for marrow

storage iron of 288 mg in

white males

and 99 mg in white females.

Some of

the

indi-vidual values must

certainly have been modified

by the diseases from which the subjects were

suf-fering.

For

example, the

highest figures in males

were

obtained in subjects with cancer, while the

lowest results were in subjects with hiatus hernia

and a history

of

previous gastrointestinal

hemor-rhage.

In spite of these qualifications, it is felt

that the results

obtained are probably

a

fair

re-flection

of

the findings in the general population,

as

the mean values in the 7 males and 9 females

suffering from uncomplicated mitral stenosis, with

no

signs of rheumatic activity, were very similar

to

those in the whole group, i.e.,

256

mg

and 70

mg,

respectively.

All these subjects had

hemo-globin levels

above 13 g per 100

ml, and only two,

both of

whom were females, had ever donated

blood

(2 and 5 pints, respectively).

The

well-marked sex difference, which is probably a

reflec-tion of the increased losses of iron via

menstru-ation and pregnancies in females, has also been

noted previously in histological assessments of

bone-marrow iron (26) and in chemical

estima-tions

of

liver iron concentrations (9, 27).

Figures

for

storage iron in Bantu

subjects

were

a

good

deal

higher

than

in

white

subjects.

The

mean

figure

of

1,629

mg in Bantu males

was

ap-proximately 6 times the value in the white

TOTAL IRON STORES IN HUMAN BONE MARROW

29-31).

The excess iron in these subjects is

largely derived from the containers used for

the

preparation of alcoholic beverages (32).

In

most

individuals, the iron is almost exclusively stored

in

the liver

and in

the

reticuloendothelial

system

of the body (8, 29).

In

an extension of

the initial

investigation,

speci-mens of bone marrow

and liver were obtained at

necropsy from white and Bantu subjects

in order

to

compare the concentrations

of

storage

iron

present in the two organs. It was of considerable

interest to find

that the concentrations of iron were

very

similar in subjects with depleted stores,

normal stores, and all degrees of iron overload.

Although it is not possible to translate the

ne-cropsy

data into accurate figures for the absolute

amounts of storage

iron present in bone marrow

and

liver, some idea of the relative capacities of

the two organs

can

be obtained from the isotopic

data.

Calculations from these results gave a mean

figure for the total marrow weight of about 1,200

g,

which is in the same range as the normal liver

weight.

This suggests that the liver and

mar-row

have a capacity to store iron that is of the

same

order.

Although the relationship probably

holds true when iron stores are normal or

moder-ately increased, there is usually significant

hepa-tomegaly when gross degrees of iron overload are

present

(3,

5),

and the liver's capacity is

there-fore

increased.

In

quantitative terms, the highest

marrow iron concentration obtained in the autopsy

study

was

8,647

_jig

per g.

If it is presumed that

the total

marrow

weight

in this subject was 1,200

g, then the

total marrow iron stores must have

been approximately 10 g. This is about half the

amount

usually present in the liver in idiopathic

hemochromatosis (3,

5).

One final

point

is worthy

of

comment.

In

the

past, the

histological assessment of hemosiderin

in the reticulum cells of marrow particles has been

widely used as possibly the best method for

as-sessing the status of iron stores in an individual

(7,

26, 33, 34).

In the present

investigation.

a

comparison was made between histological

grad-ings

and chemical estimations of storage iron in

the marrow.

The good agreement obtained over

a

wide range of iron concentrations further

con-firms the validity of the histological assessment

of

marrow

particles as a useful means of gauging

the

size of reticuloendothelial stores.

SUMMARY

Total

bone-marrow iron

stores were

estimated

in

61 human subjects by

use

of

an

isotopic

dilu-tion

technique.

The

mean

value in 24 white males

was

288

mg as

compared with 99

mg

in

17 white

females.

The high incidence of iron overload in

Bantu

adults

was

reflected in the finding of

a

mean

value of

1,629

mg

in 15 Bantu males.

Chemical analyses of iron concentrations in the

marrows

and livers of 199 white and Bantu

ne-cropsy

subjects revealed

a

close correlation

over a

wide

range

of

iron concentrations.

There

was

moderately

good agreement

between

histological

assessments

of bone-marrow

iron

stores

and chemical estimations of iron

concen-trations.

ACKNOWLEDGMENTS

The authors are indebted to Mr. L. Fatti, Mr. P. Marchand, Mr. D. Fuller, Mr. E. Joubert, Mr. G. Katz, and Mr. I. Opeskin for access to patients under their

care andto Professor B. J. P. Becker and Dr. C. Isaac-son, who kindly supplied the necropsy material. Thanks

are also due to Mr. A. D. Joffe for help with the sta-tistical analyses.

REFERENCES

1. Granick, S. Iron metabolism in animals and plants in Trace Elements. New York, Academic Press,

1958, p. 367.

2. Granick, S. Ferritin: its properties and significance for iron metabolism. Chem. Rev. 1946, 38, 379. 3. Sheldon, J. H. Haemochromatosis. London,

Ox-ford University Press, 1935, pp. 205, 208.

4. Gross, H., M. Sandberg, and 0. M. Holly. Changes in copper and iron retention in chronic diseases accompanied by secondary anemia. II. Changes in liver, spleen and stomach. Amer. J. med. Sci.

1942, 204, 201.

5. Butt, E. M., R. E. Nusbaum, T. C. Gilmour, and S. L. Didio. Trace metal patterns in diseased

states. I. Hemochromatosisand refractory anemia. Amer. J. clin. Path. 1956, 26, 225.

6. Haskins, D., A. R. Stevens, Jr., S. Finch, and C. A. Finch. Iron metabolism. Iron stores in man as

measured by phlebotomy. J. clin. Invest. 1952, 31, 543.

7. Rath, C. E., and C. A. Finch. Sternalmarrow hemo-siderin. A method for the determination of avail-able iron stores in man. J. Lab. clin. Med. 1948, 33, 81.

8. Bothwell, T. H., and B. A. Bradlow. Siderosis in the Bantu. A combined histopathological and chemi-cal study. Arch. Path. 1960, 70, 279.

9. Roth, O., B. Jasin'ski, and H. v. Bidder. Das Gewe-beeisen beim Menschen bei normalen und

logischen Zustanden. Helv. med. Acta 1951, 18,

189.

10. Mazur, A., and E. Shorr. A quantitative immuno-chemical study of ferritin and its relation to the hepatic vasodepressor material. J. biol. Chem. 1950, 182, 607.

11. Huff, R. L., P. J. Elmlinger, J. F. Garcia, J. M. Oda, M. C. Cockrell, and J. H. Lawrence. Fer-rokinetics in normal persons and in patients

hav-ing various erythropoietic disorders. J. clin. Invest.

1951, 30, 1512.

12. Huff, R. L., T. G. Hennessy, R. E. Austin, J. F. Garcia, B. M. Roberts, and J. H. Lawrence. Plasma and red cell iron turnover in normal sub-jects and in patients having various hematopoietic disorders. J. clin. Invest. 1950, 29, 1041. 13. Suit, H. D. A technique for estimating the bone

marrow cellularity in vivo using 59Fe. J. clin.

Path. 1957, 10, 267.

14. Donohue, D. M., B. W. Gabrio, and C. A. Finch. Quantitative measurement of hemopoietic cells of the marrow. J. clin. Invest. 1958, 37, 1564.

15. Donohue, D. M., R. H. Reiff, M. L. Hanson, Y. Betson, and C. A. Finch. Quantitative

measure-ment of the erythrocytic and granulocytic cells

ofthe marrowand blood. J. clin. Invest. 1958, 37,

1571.

16. Bothwell, T. H., S. Callender, B. Mallett, and L. J.

Witts. The study of erythropoiesis using tracer quantities of radioactive iron. Brit. J. Haemat. 1956, 2, 1.

17. Metz, J., N. W. Levin, and D. Hart. Effect of

al-titude on the body/venous haematocrit ratio.

Na-ture (Lond.) 1962, 194, 483.

18. Briuckmann, G., and S. G. Zondek. Iron, copper and

manganese in human organs at various ages.

Biochem. J. 1939, 33, 1845.

19. McCance, R. A., E. M. Widdowson, and L. R. B.

Shackleton. The nutritive value of fruits,

vege-tables and nuts in Medical Research Council Spe-cial Report. London, His Majesty's Stationery Office, 1936, p.21.

20. Mazur, A., and E. Shorr. Hepatorenal factors in circulatory homeostasis. IX. The identification of

the hepatic vasodepressor substance, VDM, with ferritin. J. biol. Chem. 1948, 176, 771.

21. Shoden, A., and P. Sturgeon. Hemosiderin. I. A physicochemical study. Acta haemat. (Basel) 1960, 23, 376.

22. Gillman, J., J. Mandelstam, and T. Gillman. A

com-parison of chemical and histological estimations of the iron and copper content of the livers of

Africans in relation to the pathogenesis of cyto-siderosis and cirrhosis (haemochromatosis). S.

Afr. J. med. Sci. 1945, 10, 109.

23. Dry, D. S. Improved methods for the demonstra-tion of mitochondria, glycogen, fat and iron in

animal cells. S. Afr. J. Sci. 1945, 41, 298. 24. Finch, C. A. Body iron exchange in man. J. clin.

Invest. 1959, 38, 392.

25. Kerr, L. M. H. A method for the determination of non-haem iron in bone marrow. Biochem. J.

1957, 67, 627.

26. Stevens, A. R., Jr., D. H. Coleman, and C. A. Finch. Iron metabolism: clinical evaluation of iron stores.

Ann. intern. Med. 1953, 38, 199. 27. Bothwell, T. H. Unpublished data.

28. Finch, S. C., and C. A. Finch. Idiopathic

hemo-chromatosis, an iron storage disease. A. Iron

metabolism in hemochromatosis. Medicine

(Bal-timore) 1955, 34, 381.

29. Higginson, J., T. Gerritsen, and A. R. P. Walker.

Siderosis in the Bantu of Southern Africa. Amer.

J. Path. 1953, 29, 779.

30. Wainwright, J. Siderosis in the African. S. Afr. J. Lab. clin. Med. 1957, 3, 1.

31. Bothwell, T. H., and C. Isaacson. Siderosis in the Bantu. A comparison of incidence in males and

females. Brit. med. J. 1962, 1, 522.

32. Walker, A. R. P., andU. B. Arvidsson. Iron

"over-load" in the South African Bantu. Trans. roy

Soc. trop. Med.

Hyg. 1953,

47, 536.

33. Davidson, W. M., and R. F. Jennison. The

rela-tionship between ironstorageandanaemia. J. clin. Path. 1952, 5, 281.

34. Beutler, E., W. Drennan, and M. Block. The bone marrow and liver in iron-deficiency anemia. A